Depth map generation for 3-D TV: Importance of edge and boundary information

Abstract

As an ongoing and natural development, it is reasonable to regard stereoscopic three-dimensional television (3-D TV) as the next desirable milepost beyond high-definition television (HDTV). This direction of development is evidenced by the increasing number of scientific and engineering conferences held each year on 3-D TV and related technologies. At the same time, we see that industry, academia, and standard organizations have launched major research and standardization initiatives, such as the 3-D Consortium in Japan [1], the European IST research project ATTEST [2, 3] and the JVT/MVC (MPEG- 3-D) standardization effort [4, 5]. Three-dimensional television is desirable not only because it can enhance perceived depth, but also because it promises to increase the feeling of presence and realism in television viewing [6, 7]. These attributes can heighten the entertainment value of the television medium, thereby attracting larger audiences and increasing revenues. Despite these potential benefits, the broadcasting industry has not really paid much attention to 3-D TV. One reason for such neglect is that 3-D technology is simply not mature enough to satisfy the critical conditions that are required to ensure the success of 3-D TV broadcasting.For 3-D TV to be practical and attractive to broadcasters and users alike, several critical requirements and conditions must be met. First and foremost, broadcasters need a 3-D TV approach and technology that can be easily integrated within the existing infrastructure. This condition would be best addressed by 3-D capture, transmission and display systems that are, as much as possible, backward compatible with the existing 2-D technology. It is easy to predict that the level of compatibility will be a major determinant to the adoption rate of 3-D TV services, at least in the short-term. The full implementation of these services will take several years and considerable financial resources. Therefore, it would be unreasonable to expect the broadcasting industry to quickly switch to 3-D TV unless the cost of such change is reasonably low, particularly because HDTV services are just being introduced. In fact, it would be economically beneficial if the same coding and transmission standards, as well as newly developed television sets, could be used for both 2-D and 3-D video content. For users, the main requirement is the quality of the 3-D TV service offered. In addition to providing enhanced depth sensation, realism, and sense of presence, the picture quality of the 3-D images must be as good as or better than that offered by HDTV images. On top of all these requirements, consumers need to enjoy 3-D TV that is unobtrusive and easy to use. This implies no cumbersome glasses, no limitations on viewing position, no restrictive room light conditions, and no visual discomfort. Last, but not least, there must be a rich supply of 3-D program material. In particular, at the rollout stage, it is critical that broadcasters have a sufficient amount of 3-D program content to attract viewers. © 2009 Springer-Verlag New York.